+0 Commodity PressureNo sign of generic, copyable 'AI feature' framing — highly domain-specific numerical machinery and locked execution make this hard to commoditize.
No commodity language markers presentEmphasis on Jacobian/Hessian checks and particle-weight samplersPredictive execution remains locked
+18 Model DependencyDepends on external domain data/lookups (Boeing, Copernicus) and a specialized S50 posterior; some dependencies are gated or blocked, creating brittle external coupling.
SRC BOEING EXACT-ROW LOOKUP ............. PASS / NONPREDICTIVE ONLYSRC AIRCRAFT PERFORMANCE MODEL .......... BLOCKEDS50 EVENT-SEQUENTIAL POSTERIOR ........ NOT RUN
-18 Workflow OwnershipExtremely deep, stateful workflow: checkpointed runs, public checkpoint logs, multi-branch experiments and a large frozen evaluation catalog.
Checkpointed state / Open Checkpoint / Save State UIPublic checkpoint log with datesMulti-branch numerical shakeout (16 branches / 80 cases)
-0 Distribution EmbeddednessDistribution appears local and project-specific (filesystem checkpoints, terminal UI) with no platform or channel reach shown.
Local filesystem checkpoint (C:\MH370\CURRENT_STATE)Terminal-style interface elements and system statusNo platform partners or channel markers
-8 Integration DepthStrong technical integrations with domain data (NetCDF/Copernicus), aircraft lookups, and checkpointed state — real technical entanglement with domain pipelines.
Copernicus forcing: 85 files; Forcing volume: ≈8.44 GBNetCDF forcing archiveBoeing exact-row lookup / aircraft performance source
-0 Enterprise TrustNo enterprise signals — no compliance, procurement, or customer/enterprise proof visible.
No enterprise_markers presentSite not presented as a commercial SaaS productPricing hidden; no customer proof markers
-12 Switching CostHigh switching friction due to saved state, public checkpoint history, and multi-gigabyte archived forcing data — data gravity is real here.
Checkpointed state with public logArchived forcing files (~8.44 GB)Large archival catalog (128 frozen modes; 26,952 evaluations)
-0 Monetization MaturityNo visible monetization: pricing hidden and no commercial or customer signals — appears research-first rather than productized.
Pricing visibility: hiddenCustomer proof markers: []Site described as independent modelling project, not a commercial SaaS
+4 Category BaselineVertical workflow products start safer than generic assistants.
vertical workflow
-6 Relative PlacementReduce vulnerability: specialized, stateful modeling with strong data gravity and gated execution outweighs external dependency brittleness.
Deep, stateful workflow with public checkpoint logs, checkpointed runs, multi-branch experiments and a large frozen evaluation catalog (26,952 evaluations) => high switching costs.Large domain-specific forcing archive (Copernicus NetCDF, ≈8.44 GB) and archival catalog create real data gravity that resists prompt‑replaceable attacks.Specialized numerical machinery (Jacobian/KKT/Hessian checks, particle-weight samplers, event-sequential posterior) is not expressible as simple commodity AI features.